From Public Repositories to Target Hypotheses An End-to-End Data-to-Insights Journey for scRNA and Spatial Omics with Knowledge Graphs
About the poster
Single-cell and spatial datasets hold immense potential but are hindered by curation bottlenecks and incompatible annotations. We present a scalable “Data-to-Insights Journey” that converts these sources into actionable target hypotheses through three core pillars.
1) LLM-augmented ETL, combined with atlas generation workflow, reduces processing time from days to minutes, retaining 90% accuracy. It addresses data fragmentation by unifying metadata and establishing common standards.
2) A scientist-ready browser for immediate access to AI-ready data products, for biologists to visually explore and validate datasets without coding.
3) Multiomics target ranking algorithm drives discovery by integrating single-cell data with other modalities. It synthesizes this multimodal landscape into prioritized targets. Furthermore refines the ranking with knowledge-based second line of evidence.
This end-to-end workflow accelerates the transition from raw data to reproducible insights, and has resulted in a discovery of a novel target that was later validated.
Poster was presented at Festival of Genomics and Biodata, London, 2026
